[ckan-dev] [ckan-discuss] Unable to create plug-in template

Wed Dec 8 13:25:59 UTC 2010

On 8 December 2010 11:46, Stefan Urbanek <stefan.urbanek at gmail.com> wrote:
> i'm sorry, i am still not able to create a plugin:
> $ paster create -t ckanext ckanext-mirror
> ...
> LookupError: Template by name 'ckanext' not found

Hmm, works for me. Please check that you have enabled the virtualenv
in which ckan is installed (if you are using virtualenv) - you call
post the output of paster -h and paster create --list-templates

> When I try the worker example:
> $ worker echo -d -c worker.cfg
> socket.error: [Errno 61] Connection refused
> I have full local ckan installation with working ckan environment, server
> running at http://localhost:5000.

For worker stuff you will need rabbitmq installed.

> Also I have couple of questions concerning the download/mirroring, mostly
> about desired design and desired behaviour:
> http://democracyfarm.org/f/ckan/ckan-srcmirror.png

Note that preview extension is *separate* from archiving/mirror plugin/worker.

> Who should talk to?

Myself or Friedrich.

> Few notes, findings and questions:
> - i've mirrored ckan.net DB data to localhost
> - resource "file type" is very vague, not always specified, not always
> matching with URL type

Right, intentionally for say a zip file we put format as csv if csv
inside the zip.

> - resource file type is not consistent and is mostly of three kinds: MIME,
> file extension, human readable string - i would rather not use it for
> anything but for suggesting which download/conversion plugin to try to use
> first

Agreed. Not sure though why this is relevant in first instance. For
preview stuff the work is already being done crudely in the extension
James wrote: <http://blog.ckan.org/2010/12/04/open-data-day-announcing-ckan-data-proxy/>

For archiving we aren't attempting, in first instance, to do any extraction.

> - many URLs are mostly document HTML pages, not files/resources

Those either shouldn't be there (people should just use the url
attribute) or they should have a format like index/html to indicate
they are an index/listing page. For these we really want to add a tag
of todo.scrape to the package and skip any archiving.

> - what is the desired destination for mirrored resources? ckan "local" host?
> amazon s3? another file storage?

Destination: any storage backend supported by OFS [1] -- probably
google storage at the start.

> I have also couple of design notes about how the process should look like,
> that i would like to discuss with someone with deeper ckan knowledge
> (related to http://ckan.org/ticket/698)

Happy to chat at any point -- just ping me on irc or skype.

Rufus